Abstract. We develop a model of the geographic distribution of crime in an urban area. Agents select their occupation, residence, consumption and housing. Who become criminals, the intensity and location of crime, the residence of workers and housing prices are endogenous variables affected by the level and geographic deployment of police protection. We study two opposite strategies of police deployment: concentrated protection (the police only protects an area of the city) and dispersed protection (the police is evenly deployed in all the city). In equilibrium, under concentrated protection, only rich agents are willing to pay the high housing prices in the protected area while poor agents reside in the high crime area. Under dispersed protection all areas of the city have the same income and crime levels. Comparing these equilibria, we show that crime is more likely to be higher under dispersed than concentrated protection when the skill-premium is high, dispersing the police significantly reduces its effectiveness, the proportion of income that criminal can extract from rich agents is high and the proportion they can extract from the poor is low. We also show that that concentrated protection may induce higher welfare than dispersed protection, particularly, when dispersing the police significantly reduces its effectiveness, the unprotected area is relatively large and the proportion of income that criminals can extract from the poor is low.

 

Paper jointly written with Ivan Lopez Cruz (Indiana University) and Gustavo Torrens (Indiana University).